Create supervised classification

Use the Create supervised classification block to classify images based on training data (added as areas in a collection) that you supply. Each pixel in the Area of Interest is assigned a unique class based on how similar its values are to the values of the pixels in the training data.

This block runs a cluster analysis using the training data to look at the pixels in the images and identify data properties they share. These clusters of pixels are then assigned to a class labelled according to the training data.

Algorithm

The classification methods available are:

• CART: (Classification and Regression Trees) is a form of decision tree model.
• Random Forest: a decision tree method that builds multiple decision trees and then decides which is best. See L Breiman (2001) 
• Minimum Distance: is the simplest method, it assigns each pixel to the nearest class.
• SVM: (Support Vector Machine) is a form of supervised learning model. See C Burges (1998).

Click on the cog icon to see the parameters available to refine the algorithm. 

Input data

To use a single dataset as an input, drag and drop the Use this dataset container block into the Create supervised classification block.

To add multiple datasets as inputs, use the Reuse saved dataset block.

Training data

Create a Collection on the map, and add an area for each class you would like to classify. Each area can contain multiple polygons (adding more will improve the accuracy of the classification).

Your training data collection needs to be located inside your Area of Interest.

Outputs

You can output the results of the classification as a map layer, table, or dataset.

A table of results, and an accuracy assessment are automatically generated and are shown on the Dashboard.

Learn more: How to run a classification